基于近似计算技术的FPRM逻辑功耗优化
PDF下载 (191)王一琛,王伦耀,夏银水,储著飞.基于近似计算技术的FPRM逻辑功耗优化[J].宁波大学学报(理工版),2020,33(1):45-50.DOI:
WANG Yichen,WANG Lunyao,XIA Yinshui,CHU Zhufei.Power optimization for FPRM logic using approximate computing technique[J].Journal of Ningbo University(Natural Science & Engineering Edition),2020,33(1):45-50.DOI:
| Title: | Power optimization for FPRM logic using approximate computing technique |
| 作者: | 王一琛, 王伦耀, 夏银水, 储著飞 |
| Author(s): | WANG Yichen, WANG Lunyao, XIA Yinshui, CHU Zhufei |
| 关键词: | 近似计算; 错误率; 极性搜索; 功耗优化 |
| Keywords: | approximate computation; error rate; polarity searching; power consumption optimization |
| 分类号: | TN47; TP391 |
| 文献标识码: | A |
| 摘要: | 提出了一种基于近似计算技术的FPRM逻辑功耗优化的算法, 该算法包括基于信号概率和跳变密度的固定极性Reed-Muller(Fixed Polarity Reed-Muller, FPRM)函数动态功耗模型, 基于遗传算法的以功耗优化为导向的RM逻辑极性搜索方法, 以及利用双锐积运算的RM逻辑错误率计算方法. 在错误率的约束下, 通过有选择性地删减部分乘积项, 实现功耗优化. 提出的算法用C语言实现, 并用MCNC Benchmark电路测试. 结果表明: 与原始FPRM电路功耗相比, 在平均错误率为3.21％时, 电路动态功耗平均减少了22.77％. |
| Abstract: | A power optimization algorithm based on approximate computing technique is proposed for FPRM logic circuits. The algorithm includes FPRM logic circuits dynamic power estimation model based on signal probability and transition density, a genetic algorithm for RM logic power optimization using polarity searching, and the error rate calculation for RM logic using double sharp product operation. Under the constraint of error rate, some product terms are selectively deleted to reduce power consumption. The proposed algorithm is implemented in C programming language and tested under MCNC benchmarks. The experimental results show that using the approximate computing technique presented in this work, the average dynamic power can be reduced by 22.77％ with the average error rate of 3.21％. |
| 参考文献 /References: | [1] Shin D, Gupta S K. Approximate logic synthesis for error tolerant applications[C]//Design, Automation & Test in Europe Conference & Exhibition, 2010:957-960. [2] Zou C, Qian W, Han J. DPALS: A dynamic programming-based algorithm for two-level approximate logic synthesis[C]//IEEE 11th International Conference on ASIC, 2015:1-4. [3] Ichihara H, Inaoka T, Iwagaki T, et al. Logic simplification by minterm complement for error tolerant application[C]//33rd IEEE International Conference on Computer Design, 2015:94-100. [4] Miao J, Gerstlauer A, Orshansky M. Multi-level approximate logic synthesis under general error constraints[C]//IEEE ACM International Conference on Computer-Aided Design, 2014:504-510. [5] Wu Y, Qian W. An efficient method for multi-level approximate logic synthesis under error rate constraint[C]//Proceedings of the 53rd Annual Design Automation Conference. New York, USA: ACM Press, 2016:1-6. [6] Bernasconi A, Ciriani V. 2-spp approximate synthesis for error tolerant applications[C]//17th Euromicro Conference on Digital System Design, 2014:411-418. [7] Soeken M, Gro?e D, Chandrasekharan A, et al. BDD minimization for approximate computing[C]//21st Asia and South Pacific Design Automation Conference, 2016:474-479. [8] Mozammel H A K. Design of reversible synchronous sequential circuits using pseudo Reed-Muller expressions[J]. IEEE Transactions on Very Large Scale Integration Systems, 2014, 22(11):2278-2286. [9] 尹浩凯, 马雪娇, 夏银水. 基于近似逻辑的不完全指定FPRM函数的功耗优化[J]. 宁波大学学报(理工版), 2019, 32(3):28-34. [10] 俞海珍, 汪鹏君, 张会红, 等. 基于三值多样性粒子群算法的MPRM电路综合优化[J]. 电子学报, 2017, 45(7):1601-1607. [11] Wang X, Lu Y, Zhang Y, et al. Probabilistic modeling during power estimation for mixed polarity Reed-Muller logic circuits[C]//Green Computing & Communications, IEEE, 2013:1414-1418. [12] 瞿婷, 王伦耀, 罗文强, 等. 大函数ISFPRM面积优化方法[J]. 电子学报, 2018, 46(5):1101-1106. [13] Liang J, Han J, Lombardi F. New metrics for the reliability of approximate and probabilistic adders[J]. IEEE Transactions on Computers, 2013, 62(9):1760-1771. [14] Miao J, Gerstlauer A, Orshansky M. Approximate logic synthesis under general error magnitude and frequency constraints[C]//Proceedings of the International Conference on Computer-Aided Design, 2013:779-786. [15] 王伦耀, 夏银水, 陈偕雄. 逻辑函数的双逻辑综合与优化[J]. 计算机辅助设计与图形学学报, 2012, 24(7):961-967. [16] 王伦耀, 夏银水, 陈偕雄, 等. 基于不相交乘积项的逻辑探测和拆分算法[J]. 电子学报, 2012, 40(10):2091-2096. [17] 王玉花, 王伦耀, 夏银水. 基于不相交项并行列表技术的FPRM实现[J]. 电子与信息学报, 2014, 36(9):2258-2264. |
| 备注/Memo: | 收稿日期: 2019－05－28. 宁波大学学报（理工版）网址: http://journallg.nbu.edu.cn/ 基金项目: 国家自然科学基金（U1709218, 61871242）; 浙江省自然科学基金（LY19F040004）; 宁波市自然科学基金（2019A610077）.? 第一作者: 王一琛（1993－）, 女, 浙江舟山人, 在读硕士研究生, 主要研究方向: 逻辑综合与优化. E-mail: 543746091@qq.com *通信作者: 王伦耀（1972－）, 男, 浙江宁波人, 教授, 主要研究方向: 逻辑综合与优化. E-mail: wanglunyao@nbu.edu.cn 宁波大学学报（理工版）网址：http://journallg.nbu.edu.cn/ |
